The field of the invention relates to laser based sensing systems. In particular, the field of the invention relates to a laser based optical absorption system characterized by high sensitivity, high stability, ease of use and low cost.
Sensing devices using optical absorption to measure characteristics of a sample species are well known. These devices generally use a laser to generate light that is passed through a sample gas or vapor where characteristic absorption occurs altering the intensity of the beam. A photodetector measures the intensity of the altered beam. By tuning the laser over a range of frequencies one can determine various characteristics of atoms or molecules within the species such as chemical composition, concentration, propagation direction, and velocity distribution.
A principal advantage of using optical absorption for determining such characteristics of a sample is that it is non-invasive; the devices generating and receiving the laser light do not have to be located in or necessarily near the species being measured. This can be crucial if the species, for example, is of a toxic nature or if contamination of the system is to be avoided.
In addition, optical sensing systems based on lasers are highly sensitive and selective. Such devices can accurately measure trace amounts of atomic or molecular species in a sample gas or vapor without interference of other species within the sample. This is useful for a variety of applications, including toxic gas detection, trace element detection and combustion diagnostics. In other applications such as thin film process control, detection of the velocity of species is also important, since both the velocity and concentration of a species determine the rate at which the species is deposited on a surface.
In general, these applications require detection of the species of interest with the highest sensitivity possible. The sensitivity of the system is determined by the signal to noise ratio (SNR), or the ratio of the laser absorption signal to the signal due to noise in the system. The highest sensitivity is obtained by maximizing the absorption signal from the gas while minimizing the signal due to noise in the system. For most applications the signal strength is fixed by the properties of the sample so that to increase sensitivity one must reduce the noise.
Various conventional schemes have been previously employed to reduce noise in laser absorption systems. The primary source of noise in such systems is the laser. A typical diode laser has a noise spectrum that increases rapidly at low detection frequencies. Thus, low frequency detection schemes are inherently more noisy (less sensitive) than high frequency detection schemes. This fact is well known and has led to the use of frequency modulation (FM) detection schemes (see U.S. Pat. No. 5,530,541). In FM detection the laser beam is passed through a frequency modulator which periodically shifts the laser beam frequency at a predetermined modulation frequency. The beam is then passed through the sample where the characteristic absorption takes place. The modified beam is then directed onto a detector which outputs an electrical current proportional to the intensity of the incident beam. The desired absorption information is contained in the component of the detector signal at the modulation frequency and can be extracted using a phase sensitive lock-in amplifier.
By upshifting the detection frequency from DC, where the laser noise is high to the modulation frequency, such methods greatly increase sensitivity. However, modulation frequencies typically greater than 50 MHz must be used in order to shift the detection frequency to a range where laser noise becomes negligible. Modulation frequencies of this magnitude require radio frequency (RF) electronics (oscillators, amplifiers, splitters and mixers) for signal processing. These components are complicated, expensive and susceptible to drift and other problems. In addition, the photodetector size used must be reduced linearly as the modulation frequency is increased. Modulation frequencies greater than 50 MHz require very small detectors which complicate alignment of the laser beam.
Other conventional schemes for increasing sensitivity for a laser absorption system include the method of Hobbs, U.S. Pat. No. 5,134,276. Hobbs attempts to reduce laser noise by using a noise cancellation scheme in which the laser beam is divided into a reference and a signal beam. The reference beam is directed to a reference photodetector and the signal beam is passed through the sample, where characteristic absorption takes place and is then directed to a signal photodetector. The photodetector outputs a photocurrent proportional to the intensity of incident light on the detector's surface. The reference photo-current is scaled such as to match its DC component to the DC component of the signal photo-current. The scaling is done such that substantially all components of the reference photo-current are scaled proportionally. The reference photo-current is then subtracted from the signal photo-current to produce a so-called autobalanced photo-current. The autobalanced photo-current eliminates any noise common to both reference and signal photo-currents, but retains the absorption signal down to low frequencies since it is contained in the signal photo-current alone.
This procedure reduces laser noise, however, it cannot cancel any noise that is not common to both the signal and reference beams. Shot noise is one type of such uncorrelated noise. Shot noise is a random current fluctuation that occurs when light is detected. Standard photodetectors pass current from their anode to their cathode in proportion to the number of photons incident on the detector surface. One electron is passed for each photon detected. The detection of photons, as governed by quantum mechanics, is a probalistic process. The current of the photodetector thus will be subject to random fluctuations, the size of which are determined by the standard deviation of the photon detection probability distribution. These fluctuations will be completely uncorrelated between the two photodetectors and thus cannot be canceled. In addition, any other sources of uncorrelated noise will not be canceled by the method of Hobbs. In particular any noise introduced to either beam after the laser beam is split will not be canceled.
Furthermore, conventional implementations of the Hobbs circuit in laser absorption sensing systems, as in Double beam laser absorption spectroscopy: shot noise-limited performance at baseband with a novel electronic noise canceler, Kurt L. Haller and Philip C. D. Hobbs, SPIE Vol 1435 Optical Methods for Ultrasensitive Detection and Analysis: Techniques and Applications (1991), use a diode laser without an integral grating for tuning, and direct detection (no modulation imparted on the laser beam) with the final output being derived from the log output of the Hobbs circuit. Such systems have many deficiencies, including: (a) poor spectral control, i.e. the laser wavelength is sensitive to temperature, optical feedback and other environmental effects, (b) there is no simple tuning mechanism for the laser, (c) the output is intrinsically temperature sensitive, (d) the output has a nonzero baseline, so that when the absorption is zero the absorption signal is not zero, (e) the baseline signal depends on the split ratio of the signal and reference beams, (f) scanning away from the absorption line is necessary for calibration because of the non-zero baseline signal, (g) the detection technique occurs essentially at DC, thus the system is sensitive to low frequency and acoustic noise; any low frequency noise inherent in the electronics or otherwise occurring after the splitting of the signal and reference beams will not be canceled and will reduce the sensitivity of such a conventional system, and (h) variations in the optical transmission of the signal path can only be corrected by scanning the absorption line.
Therefore, what is needed is a laser absorption system that is characterized by high sensitivity; preferably such a system should have the capability of sensing very small absorptions at the theoretical limit.
What is also needed is a laser absorption sensor system characterized by a large dynamic range, such that it has the capability to sense both small and large absorptions with no modification.
What is also needed is a laser absorption system characterized by high accuracy. Such a system would be very valuable for certain applications in which the detection and extraction of the velocity of the species are important. For example, in thin film process control measurement of the species concentration and species velocity provides species flux. The flux determines the deposition or growth rate of a film on a surface. The ability to determine flux with greater accuracy would enable a thin film to be deposited over geometrically complex surfaces with greater precision than is possible using conventional laser absorption techniques.
What is also needed is a laser absorption system characterized by low drift. That is, the output of the sensor should be time invariant when the absorption being measured is time invariant.
What is also needed is a laser absorption system with zero baseline output signal, so that when the absorption is zero, the output signal is also zero.
Additionally, what is needed is a laser absorption system with an output that is substantially independent of temperature. What is also needed is a laser absorption system with an output that is substantially independent of transmission variations through the sample region that are unrelated to absorption in the species of interest; this could occur if additional, or partially opaque, windows were placed in the beam path. Such a system would be very valuable for accurately determining the concentration of a specific species within a gas sample.
Finally, it would be desirable to achieve the foregoing objectives for a sensor system in a practical commercially viable manner. That is, the laser absorption system should be manufacturable with as few components as possible to reduce cost. Such a system ideally should have subcomponents which comprise commercially available low cost, low frequency electronics rather than high frequency specialized circuitry. In addition, such a system ideally should be rugged and capable of being operated by a user without the need for continual adjustment or recalibration.